System, Method and Device for Detection and Analysis of Wildfire Smoke

The following relates generally to wildfire risk management, and more particularly to systems, methods and devices for the detection and analysis of wildfire smoke.

Wildfires are catastrophic events that can cause significant environmental damage, economic loss and, tragically, the loss of human lives. Therefore, early detection and analysis are crucial for minimizing these impacts.

Existing methods for detecting and analyzing wildfires often lack precision, rely on data that is out of date, or do not effectively integrate multiple critical factors that contribute to wildfire risks. The absence of precise, dynamic risk assessment tools means that pre-emptive actions are not as targeted or effective as desired, leading to missed opportunities for detection, prevention and early intervention.

Accordingly, there is a need for improved systems, methods and devices for wildfire risk prediction that overcome at least some of the disadvantages of existing techniques.

There is a provided a system for detection and analysis of wildfire smoke using artificial intelligence. The system includes an imaging device configured to collect image data; a processing server for processing the collected image data, the processing server including a queueing subsystem including an initial frame assessment module configured to receive the collected image data from the imaging device; filter the collected image data based on predetermined filtration criteria; a resource allocation module configured to adjust an allocation of processing resources based on an amount of the filtered image data; a queue management module configured to prioritize the filtered image data based on predetermined priority criteria to generate prioritized image data; a motion detection subsystem including a grid cell division module configured to divide each image frame in the prioritized image data into a plurality of grid cells; a change detection module configured to determine changes between consecutive image frames of the prioritized image data by comparing pixels within each grid cell of the plurality of grid cells of the consecutive image frames; a density calculation and filtering module configured to, in response to detecting changes indicative of smoke, calculate a density of pixel changes, wherein grid cells having a density below a predetermined density threshold are discarded; and a smoke detection subsystem including a preliminary object filtering module configured to identify and remove irrelevant objects from the prioritized image data; a deep learning-based analysis module configured to analyze the prioritized image data using a smoke detection model trained to detect smoke, the smoke detection model comprising a machine-learning-based pattern detection model configured to receive the prioritized image data as an input and generate a score describing a smoke detection as an output.

The smoke detection subsystem may further include an automated camera control module configured to, in response to detecting smoke, control the imaging device to focus on an area where the smoke detection occurred.

The smoke detection subsystem may further include an alert mechanism module configured to provide an alert to relevant personnel.

The predetermined filtration criteria may include at least one of clarity, format, and integrity of the collected image data.

The processing resources may include at least one of central processing unit (CPU) resources and graphical processing unit (GPU) resources.

The predetermined priority criteria may include at least one of a higher risk of wildfire in a depicted area and signals from previously detected smoke.

The changes between consecutive image frames may include at least one of direction of smoke changes, pattern of smoke changes, and shape of smoke changes.

The score may be outputted as a single numerical score or as a categorical score.

The deep learning-based analysis module may be further configured to evaluate a predicted wildfire smoke detection against an actual wildfire occurrence to assess an accuracy of the smoke detection model.

There is provided a method for detection and analysis of wildfire smoke using artificial intelligence. The method includes receiving collected image data from an imaging device; filtering the collected image data based on predetermined filtration criteria; adjusting an allocation of processing resources based on an amount of the filtered image data; prioritizing the filtered image data based on predetermined priority criteria to generate prioritized image data; dividing each image frame in the prioritized image data into a plurality of grid cells; determining changes between consecutive image frames of the prioritized image data by comparing pixels within each grid cell of the plurality of grid cells of the consecutive image frames; in response to detecting changes indicative of smoke, calculating a density of pixel changes, wherein grid cells having a density below a predetermined density threshold are discarded; identifying and removing irrelevant objects from the prioritized image data; analyzing the prioritized image data using a smoke detection model trained to detect smoke, the smoke detection model comprising a machine-learning-based pattern detection model configured to receive the prioritized image data as an input and generate a score describing a smoke detection as an output.

The method may further include, in response to detecting smoke, controlling the imaging device to focus on an area where the smoke detection occurred.

The method may further include providing an alert to relevant personnel.

The predetermined filtration criteria may include at least one of clarity, format, and integrity of the collected image data.

The processing resources may include at least one of CPU resources and GPU resources.

The predetermined priority criteria may include at least one of a higher risk of wildfire in a depicted area and signals from previously detected smoke.

The changes between consecutive image frames may include at least one of direction of smoke changes, pattern of smoke changes, and shape of smoke changes.

The score may be outputted as a single numerical score or as a categorical score.

The method may further include evaluating a predicted wildfire smoke detection against an actual wildfire occurrence to assess an accuracy of a smoke detection model.

There is provided a device for detection and analysis of wildfire smoke using artificial intelligence. The device includes a network interface; a processor; and a non-transitory computer readable memory having stored thereon instructions that, when executed by the processor, cause the device to receive collected image data from an imaging device; filter the collected image data based on predetermined filtration criteria; adjust an allocation of processing resources based on an amount of the filtered image data; prioritize the filtered image data based on predetermined priority criteria to generate prioritized image data; divide each image frame in the prioritized image data into a plurality of grid cells; determine changes between consecutive image frames of the prioritized image data by comparing pixels within each grid cell of the plurality of grid cells of the consecutive image frames; in response to detecting changes indicative of smoke, calculate a density of pixel changes, wherein grid cells having a density below a predetermined density threshold are discarded; identify and remove irrelevant objects from the prioritized image data; analyze the prioritized image data using a smoke detection model trained to detect smoke, the smoke detection model comprising a machine-learning-based pattern detection model configured to receive the prioritized image data as an input and generate a score describing a smoke detection as an output.

The device may be further configured to, in response to detecting smoke, control the imaging device to focus on an area where the smoke detection occurred.

The device may be further configured to provide an alert to relevant personnel.

The predetermined filtration criteria may include at least one of clarity, format, and integrity of the collected image data.

The processing resources may include at least one of CPU resources and GPU resources.

The predetermined priority criteria may include at least one of a higher risk of wildfire in a depicted area and signals from previously detected smoke.

The changes between consecutive image frames may include at least one of direction of smoke changes, pattern of smoke changes, and shape of smoke changes.

The score may be outputted as a single numerical score or as a categorical score.

The device may be further configured to evaluate a predicted wildfire smoke detection against an actual wildfire occurrence to assess an accuracy of a smoke detection model.

Other aspects and features will become apparent, to those ordinarily skilled in the art, upon review of the following description of some exemplary embodiments.

Various apparatuses or processes will be described below to provide an example of each claimed embodiment. No embodiment described below limits any claimed embodiment and any claimed embodiment may cover processes or apparatuses that differ from those described below. The claimed embodiments are not limited to apparatuses or processes having all of the features of any one apparatus or process described below or to features common to multiple or all of the apparatuses described below.

As used herein, the term “about” should be read as including variation from the nominal value, for example, a +/−10% variation from the nominal value. It is to be understood that such a variation is always included in a given value provided herein, whether or not specifically referred to.

One or more systems described herein may be implemented in computer programs executing on programmable computers, each comprising at least one processor, a data storage system (including volatile and non-volatile memory and/or storage elements), at least one input device, and at least one output device. For example, and without limitation, the programmable computer may be a programmable logic unit, a mainframe computer, server and personal computer, cloud-based program or system, laptop, personal data assistants, cellular telephone, smartphone, or tablet device.

Each program is preferably implemented in a high-level procedural or object-oriented programming and/or scripting language to communicate with a computer system. However, the programs can be implemented in assembly or machine language, if desired. In any case, the language may be a compiled or interpreted language. Each such computer program is preferably stored on a storage media or a device readable by a general or special purpose programmable computer for configuring and operating the computer when the storage media or device is read by the computer to perform the procedures described herein.

A description of an embodiment with several components in communication with each other does not imply that all such components are required. On the contrary, a variety of optional components are described to illustrate the wide variety of possible embodiments of the present disclosure.

Further, although process steps, method steps, algorithms or the like may be described (in the disclosure and/or in the claims) in a sequential order, such processes, methods and algorithms may be configured to work in alternate orders. In other words, any sequence or order of steps that may be described does not necessarily indicate a requirement that the steps be performed in that order. The steps of processes described herein may be performed in any order that is practical. Further, some steps may be performed simultaneously.

When a single device or article is described herein, it will be readily apparent that more than one device/article (whether or not they cooperate) may be used in place of a single device/article. Similarly, where more than one device or article is described herein (whether or not they cooperate), it will be readily apparent that a single device / article may be used in place of the more than one device or article.

The following relates generally to wildfire risk management, and more particularly to systems, methods and devices for the detection and analysis of wildfire smoke.

Current wildfire detection models often rely on out of date or unreliable data. This lack of appropriate data makes it difficult to identify small-scale environmental changes that significantly impact wildfire detection. Moreover, many models do not holistically incorporate or integrate the fire triangle risk factor elements (weather, fuel, and potential ignition sources) alongside topographical features, which are highly valuable for accurate wildfire risk assessment. Similarly, existing models frequently offer static risk assessments that do not reflect daily changes in environmental conditions or potential ignition sources. Moreover, the presence of irrelevant objects in imagery often interferes with accurate analysis, leading to gaps in data that are highly valuable for risk prediction.

To address these challenges, embodiments disclosed herein describe techniques for the detection and analysis of wildfire smoke using artificial intelligence (AI).

Utilizing a series of interconnected subsystems, the system disclosed herein manages incoming data through a sophisticated queuing subsystem or process to handle and prioritize video frames from numerous cameras.

The system employs a motion detection subsystem, which divides frames into grid cells and identifies potential smoke movements based on pixel changes, movement direction, and shape. Candidate cells exhibiting smoke-like characteristics are further analyzed by a dedicated smoke detection algorithm. This algorithm, trained on extensive datasets, is adept at confirming the presence of smoke and enabling automated camera controls for detailed examination of suspected areas.

The system operates continuously, ensuring robust, real-time surveillance and significantly enhancing early wildfire detection capabilities. This integrated approach leverages deep learning and computer vision technologies to provide a highly efficient and accurate solution for wildfire monitoring and management.

The multi-tiered Al system for detecting and analyzing wildfire smoke includes a dynamic smoke detection model that leverages image data and machine learning algorithms. The embodiments disclosed herein are multifaceted, integrating continuous image data collection to provide a comprehensive and up-to-date tool for detection.

A queuing subsystem serves as a backbone for managing incoming video frames from hundreds of imaging devices (e.g., cameras). This subsystem efficiently monitors and allocates computing resources (central processing unit (CPU) and graphical processing unit (GPU)) based on the processing demand. The queueing subsystem queues frames meeting preliminary specifications such as format, visibility, and pixel integrity for further analysis. Other frames not meeting the preliminary specifications are discarded, optimizing resource usage and response time.

Dynamic resource scaling enables the overall system to dynamically adjust the computing resources allocated to frame processing based on the current load. This adaptability ensures efficient processing without overuse of resources, which advantageously maintains system performance during peak times such as high wind conditions, which may cause more motion in the camera's field of view.

Before queuing, each frame undergoes a light processing check to assess visibility, check format compliance, and identify any pixel damage. This initial filtering reduces unnecessary processing of unsuitable frames, thereby enhancing the efficiency of the detection process.

In the motion detector subsystem, frames are divided into grid cells, which allows the algorithm to analyze small, specific areas for changes. This localized analysis helps to accurately identify potential smoke movements without the interference of other motions in the surrounding areas.

The motion detection subsystem analyzes pixel changes between consecutive frames within each grid cell. By determining the direction and pattern of movement, the motion detection subsystem identifies upward, smoke-like motions, which are highly valuable for early smoke detection. Only movements resembling a cone-like shape, typical of rising smoke, may be considered for further analysis.

After identifying potential smoke movements, the motion detection subsystem calculates the density of changed pixels in each cell. Grid cells where the density exceeds a predefined threshold (e.g., 80% changed pixels) are marked as candidates for smoke presence. This selective filtering helps in focusing the computational effort on the most likely areas of interest.

Grid cells that pass the density threshold are selected and forwarded to the smoke detection subsystem. This staged processing ensures that only highly probable areas are analyzed in depth, improving the accuracy and efficiency of the system.

The smoke detection subsystem is configured to detect smoke within the candidate grid cells identified by the motion detection subsystem. The smoke detection subsystem employs a sophisticated algorithm trained on extensive datasets to accurately distinguish smoke from other atmospheric phenomena. Within this subsystem, an initial filtering module identifies and removes large, irrelevant objects such as lakes, clouds, buildings, and moving vehicles. Such initial filtering advantageously ensures that the motion detection subsystem focuses exclusively on areas likely to contain wildfire smoke, significantly enhancing detection accuracy by minimizing false positives.

The system features an intelligent control of camera zoom, automatically focusing on areas identified as containing potential smoke. This capability allows for detailed examination of distant smoke, enhancing detection accuracy and providing crucial time for early intervention.

The integration of these subsystems creates a robust and reliable detection mechanism, operating continuously. The seamless interaction between the queuing, motion detection, and smoke detection subsystems ensures that the system is both proactive in detecting potential fires and efficient in resource use.

Machine learning algorithms and automated data processing pipelines may be used in embodiments of the present disclosure for tasks including, without limitation, data collection, processing, and analysis, reducing manual intervention and increasing the efficiency and timeliness of risk assessments, irrelevant object detection and removal, data interpolation for weather parameters, and the generation of probabilistic risk scores.

By applying probabilistic modeling techniques to detect wildfire smoke, daily reports may be generated that indicate the likelihood of wildfire occurrences. The foregoing includes statistical analysis and modeling to quantify uncertainty and provide risk assessments in a probabilistic format.

Embodiments disclosed herein provide an innovative and dynamic tool for wildfire smoke detection, significantly improving the ability of communities, emergency services, and environmental agencies to prepare for, limit, and potentially prevent, wildfires. Advantageously, this proactive approach may minimize the devastating impacts of wildfires through early detection and accurate risk assessment, ultimately saving lives, preserving ecosystems, and reducing economic losses.

Another benefit realized from techniques disclosed herein is the ability to provide stakeholders, including fire departments, forest management agencies, and policymakers, with actionable intelligence that may guide preventive measures, resource allocation, and emergency response strategies. The foregoing may advantageously contribute to the safety of communities at risk of wildfires and the preservation of natural environments by enabling more proactive and informed wildfire management practices.

Moreover, leveraging infrastructure such as cloud computing platforms for the processing and analysis of large datasets, including imagery, weather data, and historical wildfire occurrences, further supports the computational techniques disclosed herein, enabling scalable data storage, processing, and analysis capabilities.

1 FIG. 100 Referring now to, shown therein is a block diagram of a systemfor the detection and analysis of wildfire smoke using artificial intelligence, according to an embodiment.

100 105 115 115 105 100 105 105 The systemincludes an imaging devicefor collecting image data. The image datamay be high-resolution image data. For clarity of illustration, only a single imaging deviceis shown, but it will be appreciated that the systemmay include any number of the imaging devices, e.g., a plurality of the imaging devices.

105 In an embodiment, the imaging deviceis a camera system.

105 115 115 The imaging devicecollects high-resolution imageryon a frequent basis (e.g., daily) covering a target area. The imageryincludes real-time weather data and information on topographical features and potential ignition sources (e.g., roads, campgrounds, railroads, and power lines).

100 110 115 105 174 174 120 125 130 The systemfurther includes a processing serverfor processing the image datacollected by the imaging deviceand generating output score. Such output scoreis specifically generated by a queueing subsystem, a motion detection subsystem, and a smoke detection subsystem, as will be further explained hereinbelow.

120 135 115 105 115 137 The queueing subsystemincludes an initial frame assessment moduleconfigured to receive the image datafrom the imaging device, and to filter the image databased on predetermined filtration criteria to generate filtered image data.

115 As video frames arrive from various cameras as the image data, the video frames are first assessed for basic quality metrics such as clarity, format, and integrity of the collected image data. Frames that do not meet the set criteria may be immediately discarded to ensure that only viable data consumes processing power.

120 140 137 137 135 The queueing subsystemfurther includes a resource allocation moduleconfigured to adjust an allocation of processing resources based on an amount of the filtered image data, i.e., where more filtered image datais provided by the initial frame assessment module, more processing resources may be allocated accordingly.

120 Based on the current load and the number of frames being processed, the queueing subsystemdynamically adjusts the allocation of central processing unit (CPU) and graphical processing unit (GPU) resources. This scalability is highly advantageous for maintaining optimal processing speed and efficiency, particularly during high-demand scenarios where the number of frames increases due to environmental factors (e.g., windy conditions).

120 145 137 The queueing subsystemfurther includes a queue management moduleconfigured to prioritize the filtered image databased on predetermined priority criteria.

120 137 145 147 The queueing subsystemmaintains a queue where frames that pass the initial assessment (i.e., the filtered image data) are stored temporarily. The queue management moduleprioritizes frames based on certain criteria to generate prioritized image data. Such criteria may include areas known for higher wildfire risks or signals from previously detected smoke.

125 150 147 The motion detection subsystemincludes a grid cell division moduleconfigured to divide each image frame in the prioritized image datainto a plurality of grid cells.

147 As each frame in the prioritized image datais divided into smaller grid cells, such division allows for localized analysis, focusing on specific sections of the frame where motion is detected, thereby reducing false positives from other unrelated movements.

125 155 147 The motion detection subsystemfurther includes a change detection moduleconfigured to determine changes between consecutive image frames of the prioritized image databy comparing pixels within each grid cell of the plurality of grid cells of the consecutive image frames.

125 125 The motion detection subsystemcompares pixels in consecutive frames within each grid cell to detect changes. By analyzing the direction and shape of these changes, the motion detection subsystemidentifies patterns indicative of smoke, such as upward, cone-shaped movements, e.g., according to an algorithm.

125 160 162 162 The motion detection subsystemfurther includes a density calculation and filtering moduleconfigured to, in response to detecting changes indicative of smoke, calculate a densityof pixel changes. Grid cells having a densitybelow a predetermined density threshold are discarded.

125 162 162 After detecting potential smoke-like motions, the motion detection subsystemmay use algorithms or other techniques to calculate the densityof pixel changes. Only cells with a densityabove a predetermined threshold are passed on for further analysis, ensuring that the overall system focuses on the most likely smoke indications.

130 165 147 The smoke detection subsystemincludes a preliminary object filtering moduleconfigured to identify and remove irrelevant objects from the prioritized image data.

130 The foregoing initial filtering functionality identifies and removes large irrelevant objects such as lakes, clouds, buildings, and moving vehicles, thereby enabling the smoke detection subsystemto focus exclusively on areas likely to contain wildfire smoke, significantly enhancing detection accuracy by minimizing false positives.

130 170 147 172 172 173 147 174 The smoke detection subsystemfurther includes a deep learning-based analysis moduleconfigured to analyze the prioritized image datausing a smoke detection modeltrained to detect smoke, the smoke detection modelincluding a machine-learning-based pattern detection modelconfigured to receive the prioritized image dataas an input and generate a score describing a smoke detection as the output score.

125 130 170 170 The candidate cells identified by the motion detection subsystemare further analyzed by the smoke detection subsystem, which includes a deep learning-based analysis moduleproviding a sophisticated Al algorithm trained on a large dataset of smoke images. The deep learning-based analysis moduleemploys deep learning techniques to accurately differentiate between smoke and other atmospheric phenomena.

174 174 174 174 In an embodiment, the output scoreincludes a single numerical score, with a greater output scorecorresponding to an elevated possibility of wildfire. For example, the output scoremay quantify the possibility of wildfire out of 100. Similarly, the output scoremay include a binary determination of risk level, with a value of “1” corresponding to a possibility of wildfire and with a value of “0” corresponding to no possibility of wildfire.

174 174 In an embodiment, the output scoreincludes a categorical score. For example, the output scoremay be assigned from a fixed set of three or more possible categories with each corresponding to a level of the possibility of a wildfire (e.g., none, low, medium, or high). The categorical score may be determined by converting a numerical score to a categorical score, with each category corresponding to a range of possible numerical score values.

Advantageously, embodiments disclosed herein improve wildfire smoke detection and accuracy. Techniques of the present disclosure may be used to significantly enhance the accuracy of wildfire smoke detection by leveraging high-resolution imagery, advanced machine learning algorithms, and data analysis, including a nuanced assessment of several factors.

Furthermore, the regularly (e.g., daily) updated image data for detecting wildfire smoke, reflects real-time changes in weather in this dynamic implementation. This represents a significant improvement over the static nature of conventional models, offering more timely and relevant risk assessments.

130 175 105 The smoke detection subsystemfurther includes an automated camera control moduleconfigured to, in response to detecting smoke in an area of interest, control the imaging deviceto focus on an area of interest where smoke detection occurred.

130 105 Upon confirming the presence of smoke, the smoke detection subsystemmay automatically control the imaging deviceand/or zoom functions thereof to focus in on the area of interest. This allows for detailed examination of the smoke source, aiding in the verification of fire presence and the exact location of the fire.

130 180 182 In an embodiment, the smoke detection subsystemfurther comprises an alert mechanism moduleconfigured to provide an alertto relevant personnel.

130 Once smoke is confirmed, the smoke detection subsystemtriggers an alert protocol, notifying local authorities and fire management teams. This prompt response capability is highly desirable for early intervention and potentially limiting the spread of wildfires.

In an embodiment, the predetermined filtration criteria include at least one of clarity, format, and integrity of the collected image data.

In an embodiment, processing resources include at least one of central processing unit (CPU) resources, and graphical processing unit (GPU) resources.

In an embodiment, the predetermined priority criteria include at least one of a higher risk of wildfire in a depicted area and signals from previously detected smoke.

In an embodiment, changes between consecutive image frames include at least one of direction of smoke changes, pattern of smoke changes, and shape of smoke changes.

170 172 In an embodiment, the deep learning-based analysis moduleis further configured to evaluate a predicted wildfire smoke detection against an actual wildfire occurrence to assess an accuracy of the smoke detection model.

100 172 In an embodiment, techniques presented in the systemare subject to seasonal reviews and updates to adapt to changing environmental conditions and climate patterns. This ensures that the smoke detection modelremains relevant and effective in detecting wildfire smoke year-round.

173 172 172 Furthermore, regular, seasonal reviews and updates of the predictive algorithms (e.g., the pattern detection model) of the smoke detection modeland the data provided thereto may advantageously ensure that the smoke detection modelremains accurate over time, adjusting to new patterns in climate, vegetation growth, and urban development.

Techniques disclosed herein may be used to serve as a highly valuable tool for further research and development in the field of wildfire detection and management, encouraging innovation and the adoption of advanced technologies in environmental monitoring.

2 FIG. 1 FIG. 200 200 100 Referring now to, shown therein is a flowchart of a methodfor the detection and analysis of wildfire smoke using artificial intelligence, according to an embodiment. All or parts of the methodmay be implemented at or by the systemof.

205 200 At, the methodincludes receiving collected image data from an imaging device.

In an embodiment, the imaging device is a camera system.

210 200 At, the methodfurther includes filtering the collected image data based on predetermined filtration criteria.

As video frames are received from various cameras, the video frames are first assessed for basic quality metrics such as clarity, format, and integrity of the collected image data. Frames that do not meet the set criteria may be immediately discarded to ensure that only viable data consumes processing power.

215 200 At, the methodfurther includes adjusting an allocation of processing resources based on an amount of the filtered image data.

Based on the current load and the number of frames being processed, the allocation of central processing unit (CPU) and graphical processing unit (GPU) resources is dynamically adjusted. This scalability is highly advantageous for maintaining optimal processing speed and efficiency, particularly during high-demand scenarios where the number of frames increases due to environmental factors (e.g., windy conditions).

220 200 At, the methodfurther includes prioritizing the filtered image data based on predetermined priority criteria to generate prioritized image data.

Frames are thus prioritized based on certain criteria, such as areas known for higher wildfire risks or signals from previously detected smoke.

225 200 At, the methodfurther includes dividing each image frame in the prioritized image data into a plurality of grid cells.

This division allows for localized analysis, focusing on specific sections of the frame where motion is detected, thereby reducing false positives from other unrelated movements.

230 200 At, the methodfurther includes determining changes between consecutive image frames of the prioritized image data by comparing pixels within each grid cell of the plurality of grid cells of the consecutive image frames.

200 Pixels in consecutive frames within each grid cell are compared to detect changes. By analyzing the direction and shape of these changes, the methodidentifies patterns indicative of smoke, such as upward, cone-shaped movements.

235 200 At, the methodfurther includes, in response to detecting changes indicative of smoke, calculating a density of pixel changes, and grid cells having a density below a predetermined density threshold are discarded.

Only cells with a density above a predetermined threshold are passed on for further analysis, ensuring that the overall system focuses on the most likely smoke indications.

240 200 At, the methodfurther includes identifying and removing irrelevant objects from the prioritized image data.

An initial filtering element identifies and removes large, irrelevant objects such as lakes, clouds, buildings, and moving vehicles to ensure that the focus is exclusively on areas likely to exhibit wildfire smoke, significantly enhancing detection accuracy by minimizing false positives.

245 200 At, the methodfurther includes analyzing the prioritized image data using a smoke detection model trained to detect smoke, the smoke detection model comprising a machine-learning-based pattern detection model configured to receive the prioritized image data as an input and generate a score describing a smoke detection as an output.

The candidate cells identified by the motion detector are further analyzed by a sophisticated Al algorithm trained on a large dataset of smoke images. Deep learning techniques may be used to accurately differentiate between smoke and other atmospheric phenomena.

Accordingly, machine learning algorithms may be used to analyze potential wildfire smoke data alongside current environmental conditions for optimal detection capabilities.

100 In an embodiment, the score includes a single numerical score, with a greater score corresponding to an elevated possibility of wildfire. For example, the score may quantify the possibility of wildfire out of.

In an embodiment, the score includes a binary determination of risk level, with a value of “1” corresponding to a possibility of wildfire and with a value of “0” corresponding to no possibility of wildfire.

The score may include a categorical score. For example, the score may be assigned from a fixed set of three or more possible categories with each corresponding to a possibility level of wildfire (e.g., none, low, medium, or high). The categorical score may be determined by converting a numerical score to a categorical score, with each category corresponding to a range of possible numerical score values.

250 200 In an embodiment, and at, the methodfurther includes, in response to detecting smoke, controlling the imaging device to focus on an area where the smoke detection occurred.

255 200 In an embodiment, and at, the methodfurther includes providing an alert to relevant personnel.

In an embodiment, the predetermined filtration criteria include at least one of clarity, format, and integrity of the collected image data.

In an embodiment, processing resources includes at least one of central processing unit (CPU) resources and graphical processing unit (GPU) resources.

In an embodiment, the predetermined priority criteria include at least one of a higher risk of wildfire in a depicted area and signals from previously detected smoke.

In an embodiment, changes between consecutive image frames include at least one of direction of smoke changes, pattern of smoke changes, and shape of smoke changes.

In an embodiment, the method further includes evaluating a predicted wildfire smoke detection against an actual wildfire occurrence to assess an accuracy of a smoke detection model.

200 200 In an embodiment, techniques presented in the methodare subject to seasonal reviews and updates to adapt to changing environmental conditions and climate patterns. This ensures that the methodremains relevant and effective in detecting wildfire smoke year-round.

3 FIG.A 300 Referring now to, shown therein is a devicefor the detection and analysis of wildfire smoke using artificial intelligence, according to an embodiment.

300 302 520 4 FIG. The devicemay be located at a nodeof a network, such as the networkof.

300 305 310 310 The deviceincludes a network interfaceand processing electronics, such as a processing server.

310 The processing servermay include a computer processor executing program instructions stored in memory, or other electronics components such as digital circuitry, including for example FPGAs and ASICs (not shown).

305 The network interfacemay include an optical communication interface or radio communication interface, such as a transmitter and receiver.

300 315 320 325 330 The devicefurther includes a display(e.g., an LCD screen), a sensor assembly, a power source, and a wireless antennafor wireless network communication.

300 315 310 315 310 315 User interaction with the deviceis performed through the display. The processing servermay interact with the display. Information, such as text, characters, symbols, images, icons, and other items that may be displayed or rendered on a computing device as generated by the processing servermay be displayed on the display.

320 320 320 The sensor assemblyincludes a plurality of sensors for performing different functions. For example, the sensor assemblymay include, without limitation, an imaging sensor, an air temperature sensor, a carbon dioxide sensor, a smoke sensor, an air humidity sensor, and/or a ground moisture sensor. The sensor assemblymay include further or other types of sensors in addition to or instead of the foregoing.

300 325 The devicemay be a battery-powered device and may include a battery interface for receiving one or more rechargeable batteries at the power source.

330 The wireless antennamay be used to connect to any type of wireless network, including, but not limited to, data-centric wireless networks, voice-centric wireless networks, and dual-mode networks that support both voice and data communications.

300 305 310 The devicemay include several other functional components, each of which is partially or fully implemented using the underlying network interfaceand processing server.

3 FIG.B 2 FIG. 3 FIG.A 350 200 350 350 300 Referring now to, shown therein is a schematic diagram of an electronic deviceconfigured to perform any or all of operations of the methodof., according to an embodiment. For example, a computer equipped with network function may be configured as the electronic device. The electronic devicemay be used to implement the deviceof, for example.

350 360 365 355 As shown, the electronic deviceincludes at least one processor, such as a central processing unit (CPU) and/or specialized processors such as a graphics processing unit (GPU) or other such processor unit, non-transitory computer-readable memory, and a network interface, all of which may be communicatively coupled.

350 360 365 355 According to certain embodiments, any or all of the depicted elements may be utilized, or only a subset of the elements. Further, the electronic devicemay include multiple instances of certain elements, such as multiple processors, memories, or network interfaces.

360 365 Additionally or alternatively to a processorand memory, other electronics, such as integrated circuits, may be employed for performing the required logical operations.

365 The memorymay include any type of non-transitory memory such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous DRAM (SDRAM), read-only memory (ROM), any combination of such, or the like.

365 360 In an embodiment, the memoryhas recorded thereon statements and instructions executable by the processorfor performing any of the aforementioned method operations described herein.

350 For example, in an embodiment, the electronic deviceis configured to receive collected image data from an imaging device, filter the collected image data based on predetermined filtration criteria, adjust an allocation of processing resources based on an amount of the filtered image data, prioritize the filtered image data based on predetermined priority criteria to generate prioritized image data, divide each image frame in the prioritized image data into a plurality of grid cells, determine changes between consecutive image frames of the prioritized image data by comparing pixels within each grid cell of the plurality of grid cells of the consecutive image frames, in response to detecting changes indicative of smoke, calculate a density of pixel changes, discarding grid cells having a density below a predetermined density threshold, identify and remove irrelevant objects from the prioritized image data, analyze the prioritized image data using a smoke detection model trained to detect smoke, the smoke detection model including a machine-learning-based pattern detection model configured to receive the prioritized image data as an input and generate a score describing a smoke detection as an output.

4 FIG. 500 depicts a network system, according to an embodiment.

500 512 514 514 516 516 518 518 520 512 522 522 512 300 4 FIG. 4 FIG. 4 FIG. 4 FIG. 3 FIG. The network systemincludes a serverconfigured to communicate with a plurality of imaging devices(of which one is shown atin), a plurality of database devices(of which one is shown atin), and a plurality of administrator devices(of which one is shown atin) via a network. The serveris further configured to communicate with a plurality of user devices(of which one is shown atin). The servermay be a purpose-built machine designed specifically for the detection and analysis of wildfire smoke using artificial intelligence, such as the deviceof, for example.

512 514 516 518 522 The server, imaging devices, database devices, administrator devicesand user devicesmay be, each or together, a server computer, desktop computer, notebook computer, tablet, PDA, smartphone, or another computing device.

512 514 516 518 522 520 520 The serverand/or devices,,,may include a connection with the networksuch as a wired or wireless connection to the Internet. In some cases, the networkmay include other types of computer or telecommunication networks.

512 514 516 518 522 The serverand/or devices,,,may include one or more of a memory, a secondary storage device, a processor, an input device, a display device, and an output device. Memory may include random access memory (RAM) or similar types of memory. Also, memory may store one or more applications for execution by a processor. Applications may correspond with software modules comprising computer executable instructions to perform processing for the functions described below. Secondary storage device may include a hard disk drive, floppy disk drive, CD drive, DVD drive, Blu-ray drive, or other types of non-volatile data storage.

520 512 514 516 518 522 Processor may execute applications, computer readable instructions, or programs. The applications, computer readable instructions, or programs may be stored in memory or in secondary storage or may be received from the Internet or other network. Input device may include any device for entering information into the serveror devices,,,. For example, the input device may be a keyboard, keypad, cursor-control device, touchscreen, camera, or microphone.

Display device may include any type of device for presenting visual information. For example, display device may be a computer monitor, a flat-screen display, a projector, or a display panel. Output device may include any type of device for presenting a hard copy of information, such as a printer, for example. Output device may also include other types of output devices such as speakers, for example.

512 514 516 518 522 In some cases, the serveror the devices,,,may include multiple of any one or more of processors, applications, software modules, secondary storage devices, network connections, input devices, output devices, and display devices.

512 514 516 518 522 512 514 516 518 522 512 514 516 518 522 512 514 516 518 522 Although the serveror the devices,,,are described with various components, one skilled in the art will appreciate that the serveror the devices,,,may in some cases contain fewer, additional or different components. In addition, although aspects of an implementation of the serveror the devices,,,may be described as being stored in memory, one skilled in the art will appreciate that these aspects can also be stored on or read from other types of computer program products or computer-readable media, such as secondary storage devices, including hard disks, floppy disks, CDs, or DVDs; a carrier wave from the Internet or other network; or other forms of RAM or ROM. The computer-readable media may include instructions for controlling the serveror the devices,,,and/or processor to perform a particular method.

In the present disclosure, devices, apparatus, or other components are described as performing certain acts. It will be appreciated that any one or more of these devices may perform an act automatically or in response to an interaction by a user of that device. That is, the user of the device may manipulate one or more input devices (e.g. a touchscreen, a mouse, or a button) causing the device to perform the described act. In many cases, this aspect may not be described below, but it will be understood.

512 514 516 518 522 512 522 522 522 520 As an example, it is described that the serveror the devices,,,may send information to the server. For example, a user using the user devicemay manipulate one or more input devices (e.g., a mouse and a keyboard) to interact with a user interface displayed on a display of the user device. Generally, the devicemay receive a user interface from the network(e.g., in the form of a webpage). Alternatively, or in addition, a user interface may be stored locally at a device (e.g., a cache of a webpage or a mobile application).

512 514 516 518 522 The servermay be configured to receive a plurality of information, from each of the plurality of imaging devices, database devices, administrator devices, and user devices. Generally, the information may comprise at least an identifier identifying the satellite, database, administrator, or user. For example, the information may comprise one or more of a username, e-mail address, password, or social media handle.

512 512 514 516 518 522 512 512 512 520 In response to receiving information, the servermay store the information in storage database. The storage may correspond with secondary storage of the serveror the devices,,,. Generally, the storage database may be any suitable storage device such as a hard disk drive, a solid state drive, a memory card, or a disk (e.g. CD, DVD, or Blu-ray etc.). Also, the storage database may be locally connected with the server. In some cases, the storage database may be located remotely from the serverand accessible to the serveracross a network, for example the network. In some cases, the storage database may comprise one or more storage devices located at a networked cloud storage provider.

514 516 518 522 The imaging devicemay be associated with an imaging account. Similarly, the database devicemay be associated with a database account, the administrator devicemay be associated with an administrator account, and the user devicemay be associated with a user account. Any suitable mechanism for associating a device with an account is expressly contemplated.

512 512 512 In some cases, a device may be associated with an account by sending credentials (e.g. a cookie, login, or password etc.) to the server. The servermay verify the credentials (e.g. determine that the received password matches a password associated with the account). If a device is associated with an account, the servermay consider further acts by that device to be associated with that account.

5 FIG. 1 FIG. 3 FIG. 4 FIG. 2 FIG. 1000 1000 100 300 500 1000 200 Referring now to, shown therein is a block diagram of a computing device, according to an embodiment. The computing devicemay be, for example, a component of the systemof, the deviceof, or a component of the systemof. The computing devicemay be used to implement all or part of the methodof.

1000 1020 1000 1040 1000 1060 1040 1500 The computing deviceincludes multiple components such as a processorthat controls the operations of the computing device. Communication functions, including data communications, voice communications, or both may be performed through a communication subsystem. Data received by the computing devicemay be decompressed and decrypted by a decoder. The communication subsystemmay receive messages from and send messages to a wireless network.

1500 The wireless networkmay be any type of wireless network, including, but not limited to, data-centric wireless networks, voice-centric wireless networks, and dual-mode networks that support both voice and data communications.

1000 1420 1440 The computing devicemay be a battery-powered device and as shown includes a battery interfacefor receiving one or more rechargeable batteries.

1020 1080 1110 1120 1140 1160 1180 1200 1220 1240 1260 1280 1300 1320 1340 The processoralso interacts with additional subsystems such as a Random Access Memory (RAM), a flash memory, a display(e.g., with a touch-sensitive overlayconnected to an electronic controllerthat together comprise a touch-sensitive display), an actuator assembly, one or more optional force sensors, an auxiliary input/output (I/O) subsystem, a data port, a speaker, a microphone, short-range communications systemsand other device subsystems.

1140 1020 1140 1160 1020 1180 In some embodiments, user-interaction with the graphical user interface may be performed through the touch-sensitive overlay. The processormay interact with the touch-sensitive overlayvia the electronic controller. Information, such as text, characters, symbols, images, icons, and other items that may be displayed or rendered on a computing device generated by the processormay be displayed on the touch-sensitive display.

1020 1360 1360 The processormay also interact with an accelerometer. The accelerometermay be utilized for detecting direction of gravitational forces or gravity-induced reaction forces.

1000 1380 1400 1500 1110 To identify a subscriber for network access according to the present embodiment, the computing devicemay use a Subscriber Identity Module or a Removable User Identity Module (SIM/RUIM) cardinserted into a SIM/RUIM interfacefor communication with a network (such as the wireless network). Alternatively, user identification information may be programmed into the flash memoryor performed using other techniques.

1000 1460 1480 1020 1110 1000 1500 1240 1260 1320 1340 The computing devicealso includes an operating systemand software componentsthat are executed by the processorand which may be stored in a persistent data storage device such as the flash memory. Additional applications may be loaded onto the computing devicethrough the wireless network, the auxiliary I/O subsystem, the data port, the short-range communications subsystem, or any other suitable device subsystem.

1040 1020 1020 1120 1240 1500 1040 In use, a received signal such as a text message, an e-mail message, web page download, or other data may be processed by the communication subsystemand input to the processor. The processorthen processes the received signal for output to the displayor alternatively to the auxiliary I/O subsystem. A subscriber may also compose data items, such as e-mail messages, for example, which may be transmitted over the wireless networkthrough the communication subsystem.

1000 1280 1300 For voice communications, the overall operation of the computing devicemay be similar. The speakermay output audible information converted from electrical signals, and the microphonemay convert audible information into electrical signals for processing.

While the above description provides examples of one or more systems, methods, or devices, it will be appreciated that other systems, methods, or devices may be within the scope of the claims as interpreted by one of skill in the art. Elements of each embodiment may be incorporated into other embodiments, for example, configurations discussed in relation to one embodiment may be applied to other embodiments disclosed herein. Further, it is evident that various modifications and combinations can be made without departing from the invention. The specification and drawings are, accordingly, to be regarded simply as an illustration of the invention as defined by the claims, and are contemplated to cover any and all modifications, variations, combinations or equivalents that fall within the scope of the present disclosure.